0.25
0.5
0.75
1.25
1.5
1.75
2
Empirical Game-Theoretic Analysis and the Behavior of Software Agents
Published on Jul 21, 20113762 Views
The games agents play - in markets, conflicts, or most other contexts - often defy strict game-theoretic analysis. Games may be unmanageably large (combinatorial or infinite state or action spaces), a
Related categories
Chapter list
Empirical Game-Theoretic Analysis and the Behavior of Software Agents00:00
Previously at AIPS/ICAPS…00:16
Planning in Strategic Environments01:29
Real-World Games03:06
Empirical Game-Theoretic Analysis (EGTA)06:11
EGTA Process08:17
TAC Supply Chain Mgmt Game09:15
Two-Strategy Game (Unpreempted) (1)11:52
Two-Strategy Game (Unpreempted) (2)12:57
Three-Strategy Game:Deviations14:18
Scaling #Players15:54
Improving Scalability17:44
Hierarchical Game Reduction18:29
Why Trust Reduced-Game Results?19:21
Research Questions21:30
TAC/SCM-06 Deviation Graph22:10
CDA Deviation Graph423:19
Ranking Strategies: TAC/SCM-0724:04
Strategy Ranking (TAC Travel)27:09
Strategy Ranking (CDA)27:36
DeepMaize-08 Design Exploration28:40
Iterative EGTA Process (1)30:28
Sampling Control Problem31:35
Min-Regret-First Search32:09
Min-Regret Search (1)32:51
Min-Regret Search (2)33:23
Min-Regret Search (3)33:31
Min-Regret Search (4)33:33
Min-Regret Search (5)33:55
Min-Regret Search (6)33:56
Min-Regret Search (7)34:05
Min-Regret Search (8)34:06
Min-Regret Search (9)34:07
Finding Approximate PSNE34:46
Iterative EGTA Process (2)35:09
Construct Empirical Game35:16
Payoff Function Regression35:44
Iterative EGTA Process (3)36:14
Learning New Strategies: EGTA+RL36:40
CDA Learning Problem Setup37:47
EGTA/RL Round 139:04
EGTA/RL Round 239:41
EGTA/RL Rounds 3+40:15
Strategy Exploration Problem41:35
Example42:22
FPSB2 Regret Surface43:19
Exploration Policies44:45
EGTA Applications45:19
Conclusion: EGTA Methodology46:41